Formation of C C and C O Bonds and Oxygen Removal in Reactions of Alkanediols, Alkanols, and Alkanals on Copper Catalysts

Transcription

1 pubs.acs.org/jacs Formation of C C and C O Bonds and Oxygen Removal in Reactions of Alkanediols, Alkanols, and Alkanals on Copper Catalysts María E. Sad,, Matthew Neurock, and Enrique Iglesia*, Department of Chemical Engineering, University of California at Berkeley, Berkeley, California 94720, United States Catalysis Science and Engineering Research Group (GICIC), INCAPE (UNL-CONICET), Santiago del Estero 2654, (3000) Santa Fe, Argentina Departments of Chemical Engineering and Chemistry, University of Virginia, Charlottesville, Virginia 22904, United States bs Supporting Information ABSTRACT: This study reports evidence for catalytic deoxygenation of alkanols, alkanals, and alkanediols on dispersed Cu clusters with minimal use of external H 2 and with the concurrent formation of new C C and C O bonds. These catalysts selectively remove O-atoms from these oxygenates as CO or CO 2 through decarbonylation or decarboxylation routes, respectively, that use C-atoms present within reactants or as H 2 O using H 2 added or formed in situ from CO/H 2 O mixtures via water-gas shift. Cu catalysts fully convert 1,3-propanediol to equilibrated propanol propanal intermediates that subsequently form larger oxygenates via aldol-type condensation and esterification routes without detectable involvement of the oxide supports. Propanal propanol H 2 equilibration is mediated by their chemisorption and interconversion at surfaces via C H and O H activation and propoxide intermediates. The kinetic effects of H 2, propanal, and propanol pressures on turnover rates, taken together with measured selectivities and the established chemical events for base-catalyzed condensation and esterification reactions, indicate that both reactions involve kinetically relevant bimolecular steps in which propoxide species, acting as the base, abstract the α-hydrogen in adsorbed propanal (condensation) or attack the electrophilic C-atom at its carbonyl group (esterification). These weakly held basic alkoxides render Cu surfaces able to mediate C C and C O formation reactions typically catalyzed by basic sites inherent in the catalyst, instead of provided by coadsorbed organic moieties. Turnover rates for condensation and esterification reactions decrease with increasing Cu dispersion, because low-coordination corner and edge atoms prevalent on small clusters stabilize adsorbed intermediates and increase the activation barriers for the bimolecular kinetically relevant steps required for both reactions. 1. INTRODUCTION Biomass-derived molecules typically contain excess oxygen that must be removed to render such species useful as fuels and chemicals. 1 3 Deoxygenation can proceed via elimination of H 2 OorCO 2, in which the required H and C atoms are provided either by other molecules (e.g., H 2, CO) or by the reactants molecules themselves. The low H/C ratios prevalent in biomassderived streams make hydrogen atoms a costly reactant for oxygen removal, irrespective of whether they are derived from the oxygenate molecules themselves (dehydration) or from external sources of H 2 (hydrodeoxygenation, HDO). Here, we examine alternate routes for oxygen removal, specifically those involving: (i) intramolecular use of carbon atoms through decarbonylation or decarboxylation reactions and (ii) hydrogenolysis or hydrogenation using hydrogen generated in situ via CO H 2 O reactions (water gas shift), as potential alternatives to external sources of H 2. In doing so, we have recognized the high reactivity of Cu surfaces for the formation of new C C and C O bonds via condensation and esterification reactions. Water-gas shift (WGS) 4 represents the simplest manifestation of selective removal of oxygen from an oxygenate (H 2 O) using CO as the scavenger. CO/H 2 O reactants have been used as reductants in hydrogenation reactions catalyzed by organometallic complexes, 5 such as the conversion of nitrobenzene to aniline, acetone to isopropanol, and benzaldehyde to phenol at high CO pressures in the liquid phase; such kinetic coupling between WGS and hydrogen addition reactions has not been reported, to our knowledge, on solid catalysts. Methanol and its mixtures with CO or CO H 2 O have been shown to form hydrogen-containing species that hydrogenate heavy oils using metal catalysts. 6 The in situ formation of H 2 from H 2 O has been reported in carbohydrate reforming reactions, where the H 2 formed is then used to remove O-atoms from other oxygenates. 3,7 9 We provide evidence here for deoxygenation of alkanediols, alkanols and alkanals via hydrogen formed in situ in reactions of CO with H 2 O on Cu catalysts, in pathways that combine the known properties of Cu catalysts in water gas shift 10,11 and hydrogenation-dehydrogenation reactions. Received: August 24, 2011 Published: October 25, 2011 r 2011 American Chemical Society dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

2 C C formation reactions, such as aldol-type condensations, provide attractive routes to lengthen carbon chains and to decrease O/C ratios These reactions are typically catalyzed by acids and bases and involve reactions between alkanal and alkanol intermediates to form β-hydroxy carbonyl species that subsequently dehydrate to unsaturated alkanals or alkanones. 19 On solids, aldol-type condensations require dispersed basic oxides, such as MgO, Mg-AlO x, ZnO, or alkali metal salts on silica. 20,21 Related Guerbet reactions convert primary alkanols to their β-alkylated dimers with concurrent loss of water at moderate temperatures ( K) on several metal oxides (Cu, Zn, Cr, Mo, W, Mn). 22 These coupling reactions are thought to occur via alkanol oxidation to alkanals and their subsequent condensation to form α,β unsaturated alkanals that hydrogenate to alkanols in sequential reactions. 23 Cortright et al. 8 have reported C C coupling reactions during aqueous phase reforming process on several metals supported on basic or acidic oxides. We show here that similar C C coupling reactions are catalyzed by monofunctional Cu-based catalysts from gaseous reactants at low pressures without requiring the involvement of oxide supports. These catalysts also remove O-atoms as CO x instead of H 2 O, thus preserving scarce H-atoms within products. Our results also confirm previously reported unique catalytic properties of Cu surfaces in esterification reactions 24 and provide evidence for mechanistic connections between esterification and condensation reactions. Chain lengthening involves aldol-type condensation and esterification reactions between adsorbed alkanals and basic alkoxide intermediates; these intermediates mediate the rapid equilibration between alkanals and alkanols on Cu clusters. Rates are controlled by reactions in which the alkoxide either acts as the base that abstracts the H-atom at the C α position in adsorbed alkanals (condensation) or as the nucleophile that attacks the carbonyl group in adsorbed alkanals (esterification). The lower reactivity of small Cu clusters reflects their coordinatively unsaturated surfaces, which stabilize bound reactive intermediates, thus decreasing their reactivity in the kinetically relevant steps involved in these C C and C O bond formation reactions. 2. EXPERIMENTAL METHODS 2.1. Catalyst Synthesis. Cu/ZnO/Al 2 O 3 (Cu/Zn/Al = 50/30/20) was prepared by coprecipitation from aqueous solutions (0.5 M) of Cu(NO 3 ) 3 2.5H 2 O (Aldrich, 99.99%), Zn(NO 3 ) 3 6H 2 O (Sigma-Aldrich, 98%) and Al(NO 3 ) 3 3 9H 2O (Sigma-Aldrich, 98%) with a Na 2 CO 3 solution (EMD Chemicals Inc. GR ACS) at 333 K in a stirred batch system. Mixed Cu, Zn and Al nitrate solutions were added dropwise to the stirred Na 2 CO 3 solution (0.5 M). The resulting colloids were aged by stirring for 2 h at 333 K, separated by filtration, rinsed with deionized water at 330 K (0.5 L/g solid), and treated in ambient air at 373 K overnight. These precursors were decomposed in flowing dry He (Praxair, 99.99%, 0.83 cm 3 g 1 s 1 ) by heating to 673 K (at K s 1 ) and holding for 8 h. X-ray diffractograms of precursors detected hydrotalcite structures, which decomposed upon thermal treatment to form detectable CuO and ZnO phases (Supporting Information, Section S1). X-ray diffraction (XRD) patterns were measured using a Siemens D5000 diffractionmeter and Cu Kα radiation (2θ =5 80 ; 2 min 1 ). Cu-free ZnO-Al 2 O 3 was prepared using the same procedure and Zn/Al ratio (ZnO/Al 2 O 3 ;Zn/Al= 60/40, S BET =65m 2 g 1 ) as for Cu/ZnO/Al 2 O 3 to assess the reactivity of these oxides in alkanediol and alkanol reactions. Cu clusters dispersed on carbon (HSAG 300, TIMREX, 230 m 2 g 1 ) and SiO 2 (CAB-O-SIL HS-5, 311 m 2 g 1 ) were prepared by their respective impregnations to incipient wetness with Cu(NO 3 ) 3 2.5H 2 O solutions. Several SiO 2 -supported catalysts containing 5 20% wt. Cu were prepared by incipient wetness impregnation changing the temperature of thermal treatment (from 548 to 923 K) and using triethanolamine (Fluka, >98.5%) additives (Cu/triethanolamine molar ratio was 1:1) to vary Cu dispersion and cluster size. 25 These catalysts were treated in ambient air at 383 K for 12 h and then in flowing dry air (0.83 cm 3 g 1 s 1 ) by heating to 773 K (0.083 K s 1 ) and holding for 5 h, unless other treatment protocols are specifically indicated. Samples were then cooled to ambient temperature and treated in flowing 10% H 2 /He (Praxair, %, 5.56 cm 3 g 1 s 1 ) by heating to 553 K (at K s 1 ) and holding for 2 h and ultimately exposed to 1% O 2 /He (Praxair, %, 0.83 cm 3 g 1 s 1 ) at ambient temperature for 1 h to passivate them before exposure to ambient air Surface Area, Metal Dispersion and Temperatureprogram Reduction Measurements. BET surface areas were measured using a Quantasorb 6 Surface Analyzer (Quantachrome Corp.) and N 2 at its normal boiling point after treating samples in 0.1 Pa dynamic vacuum at 473 K overnight. The Cu dispersion, defined as the ratio of surface (Cu 0 ) s to total Cu atoms, was determined by titration of surface Cu atoms with N 2 O at 313 K using a stoichiometry of N 2 O/(Cu 0 ) s of Cu s þ N 2 O f Cu s O Cu s þ N 2 on prereduced samples (0.3 g) treated in 20% H 2 /Ar (Praxair, %, 5.6 cm 3 g 1 s 1 ) by heating to 553 K (at K s 1 ) and holding for 1 h. Samples were then flushed with Ar (Praxair, %, 5.6 cm 3 g 1 s 1 ) at 553 K for 0.5 h, cooled to 313 K, and exposed to flowing 0.5% N 2 O/Ar (Praxair, %, 5.6 cm 3 g 1 s 1 ). N 2 O uptakes were measured from the intensity of N 2 O ions (44 amu) (Inficon, Transpector Mass Spectrometer) in the inlet and outlet streams. Mean Cu cluster diameters were estimated from Cu dispersions by assuming hemispherical clusters with the atomic density of bulk Cu. 27 The reduction temperature of supported Cu catalysts was measured using temperature programmed reduction (TPR) methods. Samples (0.4 g, μ) were placed within a quartz reactor, heated in Ar (Praxair, %, 5.6 cm 3 g 1 s 1 ) to 673 K (Cu/ZnO/Al 2 O 3 )or 773 K (Cu/SiO 2 ) at 0.33 K/s and held at this temperature for 0.5 h in order to remove H 2 O before reduction. The samples were cooled to ambient temperature in Ar, the gas stream was switched to 20% H 2 /Ar (0.67 cm 3 s 1, Matheson UHP, certified mixture) and the temperature was then increased to 1000 at K s 1. The effluent was analyzed by a quadrupole mass spectrometer (Inficon, Transpector Mass Spectrometer) Catalytic Turnover Rates and Selectivities. Catalytic rates were measured with gas phase reactants at 503 K in a packedbed with plug-flow hydrodynamics using prereduced and passivated catalysts ( μm, g) diluted with SiO 2 ( μm aggregates, Fluka, purum. P.a., treated at ambient temperature with 0.5 M HNO 3 and treated in dry air at 673 K). All samples were treated again in 5% H 2 /He (Praxair, %) at 533 K ( K s 1 ) for 1 h before catalytic measurements. Liquid reactants (1,3-propanediol, Aldrich, 99.6%; propanol, Sigma-Aldrich, 99.7%; propanal, Acros, >99%; deionized water) were introduced using a syringe pump (Cole Parmer, series) by vaporizing each reactant into flowing He (Praxair, UHP) at 363 K (propanol, propanal), 378 K (water), or 433 K (1,3-propanediol). CO (Praxair, %) and H 2 (Praxair, %) were also used as reactants. Reactant and product concentrations were measured by gas chromatography (Agilent 6890) using a methyl silicone capillary column (Agilent, HP-1, 50 m, 0.32 mm ID, 1.05 μm film) connected to a flame ionization detector and a packed column (Porapak Q, 4.8 m, 1/8 00 diameter, ) connected to a thermal conductivity detector. Mass selective detection (HP 5972) after chromatographic separation (HP 5890 GC) was used for speciation of all reaction products dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

3 Table 1. 1,3-Propanediol Conversion and Product Formation Rates and Selectivities on Cu/ZnO/Al 2 O 3 a pressure [kpa] selectivity b [%] Formation rates [10 8 moles g 1 s 1 ] H 2 CO H 2 O conversion [%] propene propanal propanol propene propanal propanol a 503 K, 216 g catalyst-ks (mol 1,3-propanediol) 1 residence time, 0.8 kpa 1,3-propanediol, balance He, n.d.< 0.05%moles. b Carbons based selectivity, small amounts of acrolein, C 5 or C 6 also detected. Scheme 1. Reaction Network for 1,3-Propanediol Deoxygenation Figure 1. Effect of H 2 pressure (from added H 2 or H 2 formed in situ via water gas shift) on propanol propanal molar ratio during 1,3-propanediol reactions on Cu/ZnO/Al 2 O 3. Alkanediol H 2 reaction (0), alkanediol H 2 CO reaction (O), alkanediol CO H 2 O reaction (2), approach to equilibrium for propanal propanol H 2 reaction ([),----equilibrium line. Propanol propanal ratio for low H 2 pressures are shown in the small plot [216 g catalyst-ks (mol 1,3- propanediol) 1 residence time, 503 K, 0.8 kpa 1,3-propanediol, balance He]. a Approach to equilibrium for propanal propanol interconversion: η 1 =[P propanol ]/[P H2 ] [P propanol ] 1/K 1 eq, K 1 eq : equilibrium constant for propanal propanol interconversion at 503 K RESULTS AND DISCUSSION ,3-Propanediol Reactions with H 2 and CO H 2 O Coreactants. Table 1 shows conversion, selectivities, and rates for 1,3-propanediol reactions on Cu/ZnO/Al 2 O 3 with and without added H 2 and CO H 2 O as coreactants. Products were not detected on SiO 2 (used as diluent) or in empty reactors; thus gas phase and SiO 2 -mediated reactions do not occur at these conditions. ZnO/Al 2 O 3 supports showed some propanediol conversion to acrolein, as discussed below. 1,3-Propanediol reactants were completely converted to equilibrated propanal propanol mixtures (Figure 1, thermodynamic data 28 ) and to trace amounts of other products on Cu/ZnO/ Al 2 O 3, even without H 2 coreactants (Table 1, first row). Propanal propanol equilibriation was confirmed at all conditions from approach to equilibrium parameters (η 1,defined in Figure 1) that were unity within experimental accuracy (Figure 1). Propene, allyl alcohol (2-propen-1-ol), acrolein (2-propenal) and C 5 C 6 oxygenates were also detected at low concentrations in the effluent stream. Propanal was the predominant product in the a Equilibrium constants calculated at 503 K, 28 thermodynamic properties of 3-hydroxy-propanal were estimated by group contribution method in Aspen v7.1 absence of added H 2 (86% selectivity) and small amounts of acrolein were also formed, consistent with thermodynamic expectations. 28 These products are consistent with the 1,3-propanediol conversion pathways shown in Scheme 1, in which allyl alcohol forms initially (thermodynamically favored, K eq = bar at 503 K 28 ) and then undergoes dehydrogenation to produce acrolein with unfavorable thermodynamics (K eq = 0.04 bar at 503 K 28 ). The double-bond isomerization of allyl alcohol to propanal occurs rapidly via intramolecular or intermolecular hydrogen transfer and is favored by thermodynamics (K eq = at 503 K 28 ). Fast hydrogenation steps then form propanol in amounts dictated by thermodynamics at the prevalent H 2 pressures. Propene and propane can also form via dehydration and hydrogenation reactions, but usually require acid sites; propane was not detected with propanediol H 2 reactants and propene selectivities were very small (<1%, Table 1). 1,3-Propanediol forms propanal on Cu/Al 2 O 3 at 523 K 29 via pathways proposed to involve 3-hydroxypropanal dehydration to acrolein and subsequent hydrogenation. As a result, Scheme 1 also includes this route. The prevalent equilibration among these products in these experiments precludes more specific conclusions about the actual routes involved in the equilibration of propanediol reactants with propanol propanal products. H 2 -propanediol reactant mixtures predominantly formed equilibrated propanal propanol mixtures with traces of allyl dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

4 alcohol, acrolein, propene, and C 5 C 6 oxygenates (Table 1). Propanol propanal ratios were proportional to H 2 pressures and their values were accurately described by the thermodynamics of their interconversion 28 (10 80 kpa H 2, Figure 1). Propane was not detected among reaction products, even at residence times that equilibrated propanol and propanal products, indicating that propanol hydrogenolysis and propene hydrogenation do not occur at detectable rates at these conditions. 24,30,31 Figure2showsconversionsandproductyields(molesofCini product per mole of C in 1,3-propanediol in the feed) as a function of residence time for reactions of 1,3-propanediol with H 2 (10 kpa) as coreactant. The equilibrated nature of propanal propanol interconversions allow us to consider these two molecules as a kinetically indistinguishable chemical lump, denoted hereinafter as the reactant pool. The nonzero initial slopes for propanal propanol and acrolein yields (Figure 2) indicate that they form either directly from propanediol or via sequential routes mediated by reactive intermediates present at trace concentrations (e.g., allyl alcohol or 3-hydroxypropanal). Acrolein yields increased with residence time and then decreased, consistent with the involvement ofacroleinasanintermediateintheformationofpropanoland propanal. Propene yields show a zero initial slope, consistent with its formation via slow reactions of equilibrated propanol propanal mixtures and not directly from propanediol (Scheme 1). The possible involvement of the support in 1,3-propanediol conversion was examined by comparing rates with 1,3-propanediol H 2 and 1,3-propanediol He on Cu/ZnO/Al 2 O 3 and ZnO/ Figure 2. Effect of residence time on 1,3-propanediol conversion and product yields on Cu/ZnO/Al 2 O 3 (503 K, 0.8 kpa 1,3-propanediol, 10 kpa H 2, balance He). Al 2 O 3 (Table 2). 1,3-Propanediol conversion rates were much larger on Cu/ZnO/Al 2 O 3 than on ZnO/Al 2 O 3 (40 and 60 times, with and without H 2 ) (Table 2); these effects of Cu are even stronger than indicated by these data when we consider the smaller kinetic driving force imposed by the higher conversions prevalent on the Cu/ZnO/Al 2 O 3 sample. Acrolein was the main product on ZnO/Al 2 O 3 without added H 2 and acrolein and propanal were not in equilibrium on Cu/ZnO/Al 2 O 3 or ZnO/ Al 2 O 3 at any H 2 pressure. On Cu/ZnO/Al 2 O 3, the approach to equilibrium parameter (η 2, defined in Table 2) was 10, indicating that propanal did not form from acrolein; we conclude that propanal forms via either isomerization of allyl alcohol or hydrogenation of bound acrolein species before their desorption. On ZnO/Al 2 O 3, the acrolein-propanal approach to equilibrium factor was (Table 2), suggesting that propanal forms slowly from allyl alcohol or acrolein and that acrolein is the initial product of propanediol reactions in the absence of Cu. The approach to equilibrium values for propanal propanol H 2 reactions (η 1 ) were near unity on Cu/ZnO/Al 2 O 3 but larger than unity on ZnO/Al 2 O 3 (Table 2), consistent with hydrogenation of propanal to propanol on Cu sites and with propanal as an earlier product of propanediol reactions than propanol on ZnO/Al 2 O 3 catalysts. The presence of H 2 (10 kpa) did not influence diol conversion rates on Cu/ZnO/Al 2 O 3 (Table 2), but increased diol conversion rates on ZnO/Al 2 O 3 (Table 2) by hydrogenating acrolein to propanal and overcoming, in this manner, the thermodynamic constraints associated with acrolein formation. We conclude that ZnO/Al 2 O 3 supports catalyze 1,3- propanediol dehydration-dehydrogenation to acrolein, but that sequential acrolein hydrogenation to propanal and propanol and hydrogen transfer isomerization of allyl alcohol to propanal occur slowly in the absence of Cu. In the presence of Cu, propanal forms rapidly via diol dehydration to allyl alcohol and subsequent hydrogen transfer reactions to form equilibrated propanal propanol mixtures. Table 1 shows the effects of the H 2 molecules formed in situ via water gas shift when CO H 2 O mixtures were introduced together with 1,3-propanediol on Cu/ZnO/Al 2 O 3. WGS rates without diol coreactants are shown in Table 3. CO conversions decreased with time on stream from 24% (0.9 ks) to 4% (after 10.8 ks) with 21 kpa H 2 O at 503 K (and from 60 to 15% with 80 kpa H 2 O); deactivation was not observed, even after 25 ks, however, when H 2 (30 kpa) was added with CO/H 2 O reactants (Table 3), as also shown in previous studies. 32 H 2 was added to all reactant streams containing CO and H 2 O to prevent deactivation processes that would otherwise interfere with accurate rate and selectivity measurements. H 2 effluent concentrations decreased when 1,3-propanediol (0.8 kpa) was added to CO H 2 O Table 2. Effect of Support on 1,3-Propanediol Conversion and Product Formation Rates and Selectivities on Cu/ZnO/Al 2 O 3 a carbon-based selectivity [%] product formation rates d [10 8 moles g 1 s 1 ] catalyst H 2 [kpa] conversion [%] η 1 b η 2 c acrolein propanal propanol acrolein propanal propanol sum e Cu/ZnO/Al 2 O e Cu/ZnO/Al 2 O f ZnO/Al 2 O n.d n.d. 110 f ZnO/Al 2 O a 503 K, 0.8 kpa 1,3-propanediol, balance He; n.d.< 0.05%moles. b Approach to equilibrium for propanal propanol H 2 reaction defined in Figure 1. c Approach to equilibrium for acrolein-propanal H 2 reaction η 2 =[P propanol ]/[P H2 ] [P acrolein ] 1/K eq. d Integral rates, calculated by dividing the 1,3- propanediol fractional conversion by the residence time. e 10.8 g ks mol 1 residence time. f 216 g ks mol 1 residence time dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

5 Table 3. H 2 Formation Rates during CO H 2 O Reactions on Cu/ZnO/Al 2 O 3 in the Presence or Absence of 1,3-Propanediol Coreactants a pressure b [kpa] CO conversion c [%] net H 2 formation rate d [10 6 moles g 1 s 1 ] H 2 consumption rate e [10 6 moles g 1 s 1 ] time on stream [ks] time on stream [ks] time on stream [ks] CO H 2 O H 2 1,3-PPD f f a 503 K, 19.7 g ks mol CO 1 residence time, balance He. b Reactor inlet pressure. c CO conversion calculated as (moles CO out -moles CO in )/mols CO in. d Calculated as: moles H 2 produced/residence time expressed as g catalyst-g (moles CO in ). e Calculated as: moles propanol produced from 1,3PPD/ residence time expressed as g catalyst-g (moles 1,3-PPD in ). Propanol was used because it was the only product that consumes H 2 (1 mol of H 2 is necessary to produce 1 mol of propanol) produced from the alkanediol. f 64.9 g catalyst-ks (mol CO) 1 residence time. Scheme 2. Reactions Steps Involved in Propanol Propanal Equilibration and C C and C O Coupling Reactions from C 3 Intermediates on Cu Catalysts inlet streams, either because of WGS inhibition by diols or H 2 depletion by conversion of 1,3-propanediol to propanol. H 2 consumption rates via the latter route were much smaller than the decrease in H 2 effluent molar rates observed upon diol addition (Table 3), indicating that 1,3-propanediol (or propanol propanal products) inhibits WGS reactions, possibly via competitive adsorption or reactions with CO and H 2 O. 1,3-Propanediol conversions and selectivities with CO/H 2 O and CO/H 2 O/H 2 coreactants are shown in Table 1. Diol conversions were >95% and equilibrated propanol propanal mixtures were the main products. Propanol propanal interconversion involves the formation of adsorbed alkoxides, via O H activation in alkanols or H-addition to alkanals, 33 in steps that are fast and quasi-equilibrated at all experimental conditions. Scheme 2 shows that ROH species can undergo O H activation on Cu surfaces to form alkoxides. These alkoxides can then dehydrogenate to form alkanals (as shown by propanol propanal equilibration). These adsorbed alkanals can subsequently react with vicinal alkoxides or C 3 H 5 O* species formed via alkanal dehydrogenation, mediated by bound alkoxides to form C O orc C bonds via bimolecular esterification or condensation reactions, respectively (as discussed in section 3.2.1). The C 3 H 5 O* nomenclature is used to indicate that while the key surface intermediates for both esterification and condensation contain the same atomic composition, their atomic structures can be different. The low alkane selectivities observed in alkanol H 2 or alkanediol H 2 reactions indicate that Cu does not activate the C O bonds in these oxygenates before formation of new C C orc O bonds and that hydrogenolysis of these oxygenates therefore does not occur. We conclude from these data that 1,3-propanediol is completely converted to equilibrated propanol propanal mixtures even at short residence times and modest temperatures ( 500 K) on Cu/ZnO/Al 2 O 3 (Table 1). The fast nature of the 1,3-propanediol dehydration and hydrogenation-dehydrogenation on all Cu catalysts tested (also on Cu/SiO 2 catalysts, data not shown) preclude a more rigorous kinetic and mechanistic analysis of these reactions. The rapid formation of equilibrated propanal propanol led us to examine the identity and mechanism of subsequent reactions of these molecules at longer residence times in the presence of H 2, CO, H 2 O and mixtures thereof Catalytic Reactions of Propanol and Propanal on Cubased Catalysts. In a later section (3.2.2), we show that supports do not influence rates and selectivities in reactions of propanal propanol H 2 mixtures. In view of this, we focus here on Cu/SiO 2 catalysts, which can be prepared with a wide range of mean Cu cluster sizes, in our kinetic and mechanistic studies. As in the case of 1,3-propanediol, propanol, propanal, and their mixtures equilibrated rapidly at all H 2 pressures (Figure 3, line denotes equilibrium ratios 28 ) and also when H 2 was formed in situ via CO/H 2 O reactions (Figure 3). As stated earlier, propanol and propanal can be rigorously treated as a single chemical species (reactant pool) and conversions and selectivities are based on the inlet and outlet molar rates of these combined reactants Primary and Secondary Products in Propanol Propanal H 2 Reactions on Cu Catalysts. Equilibrated propanol propanal H 2 mixtures formed a broad range of products on 5% wt. Cu/SiO 2 catalyst (5.6% dispersion; 0.64 kpa propanol, 80 kpa H 2 ; Figure 4) at residence times (2160 g-ks/(mol propanol) 1 ) significantly larger than those required to form these equilibrated mixtures from 1,3-propanediol (216 g-ks/(mol 1,3-propanediol) 1 ). These products include oxygenates, predominantly 3-pentanone and propyl propionate, with 2-methylpentanal and 2-methyl-3-pentanone as minority species. Propene and propane were detected in small concentrations and propane/propene molar ratios were smaller than predicted from their interconversion equilibrium ( 0.1, K eq = at 503K 28 ), suggesting that small amounts of propene formed via dehydration of propanol undergo slow secondary hydrogenation-dehydration on Cu. 34 The trace amounts of 3-pentanol detected correspond to its unfavorable equilibrium with 3-pentanone (K eq = bar 1 at 503K 28 ), as also found for the other C 6 alkanals and alkanones and their respective alkanols dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

6 Scheme 3. Reaction Network for Propanol Propanal Reactions on Cu-based Catalyst Figure 3. Effect of H 2 pressure (from added H 2 or H 2 formed in situ via water gas shift) over propanol propanal molar ratio during propanol H 2 or propanal H 2 reactions on Cu/ZnO/Al 2 O 3.Propanol H 2 (2), propanal H 2 (O), propanol CO H 2 O reaction: 21 kpa H 2 O, 8 kpa CO (9), 30 kpa H 2 O, 30 kpa H 2, 10 kpa CO (0),--- propanol propanal equilibrium line [503 K, 2160 g catalyst-ks (mol 1,3-propanediol) 1 residence time, 0.64 kpa propanol or propanal, balance He]. Figure 4. Selectivity b vs conversion c for propanol propanal H 2 reactions and approach to equilibrium for propanol to propanal transformation (η 1 ) on 5 wt % Cu/SiO 2. Selectivities to propane (b), propene (0), 3-pentanone (w), propyl propionate (2), 2- methyl-3- pentanone ()), 2-methyl pentanal (1) and approach to equilibrium for propanol to propanal and H 2 reaction, η 1 (O) [5% wt. Cu/SiO 2, 5.6% dispersion, 503 K, 2160 g catalyst-ks (mol propanol) 1, 0.64 kpa propanol, 80 kpa H 2, balance He]. a Approach to equilibrium for propanal propanol interconversion defined in Figure 1. b Defined as: S i =(F out in out i )/(F propanal F propanal + propanal ) where F j are the molar rates. c in Defined as: X =(F propanol the molar rates. out F propanal + propanal in )/(F propanal ) where F j are The effects of residence time (and pool conversion) on selectivities were used to probe the primary or secondary nature of all products formed (Figure 4). The approach to equilibrium (η 1 ) for propanol/propanal interconversion was near unity at all conversions and residence times. 3-Pentanone, propyl propionate, 2-methyl pentanal, 2-methyl 3-pentanone and propene selectivities gave nonzero values when extrapolated to zero conversion, consistent with their direct formation as primary products from the propanal propanol reactant pool; propane gave zero initial slope, as expected from its formation via secondary hydrogenation of propene derived from propanol dehydration instead of via primary hydrogenolysis reactions of propanol. Equilibrated propanal propanol H 2 mixtures preferentially formed C 5 and C 6 molecules with fewer O-atoms via condensation and esterification reactions but only small amounts of deoxygenated C 3 molecules. The formation of each product is described schematically in Scheme 3 as four general types of reactions from equilibrated propanol propanal mixtures: (i) aldol-type condensation reactions 20 by activation of C H bonds at α-positions to form an aldol intermediate that can further react to 2-methyl-3-pentanone and 2-methylpentanal or undergo decarbonylation or decarboxylation to form 3-pentanone (ii) formation of linear C 6 alkanals and alkanones by activation of C H bonds at β-positions (iii) esterification reactions to form propyl propionate 35 (iv) dehydration of propanol to propene and subsequent hydrogenation Base-catalyzed aldol condensation involves reactions of carbonyl species with enolate intermediates of another carbonyl species to form β-hydroxy carbonyl compounds (aldols), which subsequently dehydrate to α,β-unsaturated carbonyls and hydrogenate to the corresponding alkanal or alkanone. 19,36 On bases, low hydrogenation rates (or the absence of H 2 )lead to α,β-unsaturated species and condensation occurs only for carbonyl species that contain an α-h atom. For example, propanal reactions on Mg/Al oxides form aldols as primary products; these species undergo fast dehydration-hydrogenation or hydrogenolysis to form 2-methyl 3-pentanone and 2-methyl-pentanal. Cu-based catalysts form 2-methyl-pentanal and 2-methyl-3- pentanone (Figure 4, Scheme 3), the expected products of aldoltype condensation reactions of propanal propanol mixtures, but 3-pentanone is the predominant product, apparently as a result of decarbonylation or decarboxylation (instead of dehydration) of aldol-type intermediates, in contrast with the data reported for Cu supported on basic solids, 31 for which C 6 alkanones are the dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

7 main products. Aldol-type reactions of alkanols are known as Guerbet reactions 37 and occur on oxides of Cu, Zn, Cr, Mo, W, and Mn 21 with loss of oxygen as H 2 O instead of CO x. 1-Alkanols with n-carbons react on Cu/ZnO/Al 2 O 3 to form esters with 2n carbons and alkanones with 2n or 2n 1 carbons (2n 1 alkanones form by CO x loss) via steps that are thought to involve aldol-type species formed via recombinative desorption of the H-atoms that result from C H activation steps on basic support sites on Cu surfaces, 30 as also proposed on Cr oxide catalysts. 38 These studies have implicated the essential involvement of support basic sites in C C bond formation. We show in the present paper (section ), however, that monofunctional Cu surfaces catalyze aldol-type C C coupling reactions and the removal of O-atoms using C-atoms within oxygenate molecules without any detectable assistance by the support. Esters are typically formed via reactions of organic acids with alkanols using acid catalysts 19 but can also form from alkanals lacking an α-hydrogen via Tishchenko or Cannizaro reactions using bases as catalysts. The Tishchenko route 35,39 involves reactions of alkanals with alkoxide species, while the Cannizaro reaction 40 involves the reduction of an alkanal species to the alkanol, the oxidation of another alkanal molecule to carboxylic acid, and their subsequent bimolecular reactions. Another route proposed for ester formation involves hemiacetal intermediates formed via reactions between alkanals and alkanols on CuO/ ZnO/Al 2 O 3. 24,31 Next, we provide evidence that condensation and esterification proceed via monofunctional routes on Cu/SiO 2 (Figure 4) and that metallic Cu cluster surfaces stabilize all intermediates required to equilibrate alkanal-alkanol mixtures and to mediate their C CandC O formation reactions without assistance by SiO 2 supports. Cu clusters dissociate O Hbondsinalkanolsto form surface alkoxides 33 that dehydrogenate to alkanals in the steps shown in Scheme 2. Adsorbed alkoxides also act as bases in reactions with alkanals via the two routes also shown in Scheme 2. In one route, alkoxides abstract the acidic α-hydrogen in adsorbed alkanals to form adsorbed enolate species that ultimately attack the electrophilic carbonyl group in a vicinal alkanal to form a new C C bond in aldol-type condensation steps. The basic O-atom in adsorbed alkoxides can also attack the electrophilic carbonyl group in a vicinal alkanal to form a new C O bond and the esterification products. Linear alkanal (hexanal, 3-hexanone) or alkanol products were not detected, indicating that C H bonds at β-positions are unreactive, as typically observed in aldol condensations by acids or bases. 36 We discuss both of these paths in more detail below in the specific context of the kinetic response of condensation and esterification reactions to the concentrations of propanal, propanol, and H Effects of Support in Propanol Propanal H 2 Reactions on Cu. Figure 5 shows the selectivity to the various products formed from propanol propanal H 2 reactants on Cu clusters supported on SiO 2, carbon, and ZnO/Al 2 O 3 at similar reactant pool conversions ( 5 7%). All catalysts predominantly formed propyl propionate and 3-pentanone. Additional SiO 2 (0.3 g), present as a physical mixture with 20% wt. Cu/SiO 2 (0.05 g), did not influence rates or selectivities. We conclude from the prevalence of esterification and condensation products on all Cu clusters, irrespective of the amount or identity of the support that condensation and esterification, occur via monofunctional Cu-catalyzed pathways. Although it has been reported that esterification can occur on Cu surfaces, 21 the ability Figure 5. Products selectivity (defined in Figure 4) for propanol/ propanal/h 2 reaction on Cu supported catalysts (pool conversion = 5 7%). Selectivity to 3-pentanone (black) and propyl propionate (gray). 503 K, 0.64 kpa propanol, 10 kpa H 2, balance He, 2700 g-ks (mol propanol) 1 residence time for 10% wt. Cu/SiO 2, 20% wt. Cu/ SiO 2 and 10% wt. Cu/C, 540 g-ks (mol propanol) 1 residence time for 50% wt. Cu/ZnO/Al 2 O 3. of Cu metal surfaces to catalyze condensation reactions by activating the C H bond of the α C-atom (C α H) without assistance from supports appears not to have been previously recognized; such properties seem plausible, however, in light of the known ability of Cu surfaces to activate C H, C O and O H bonds 41 in oxygenates and to form adsorbed alkoxides that can act as basic moieties in condensation and esterification catalytic sequences. Intramolecular C-atoms (instead of H-atoms) are used to selectively remove oxygen from aldol species on all catalysts to form 3-pentanone, without requiring the presence of a support with basic properties. Decarbonylation was also observed by other authors 15 yielding n-pentane and CO from 2-methyl 2-pentenal using Pt and Pd catalyst but not on Cu surfaces. In section 3.2.4, we also show that the rates of formation of both propyl propionate and 3-pentanone are proportional to the number of Cu atoms exposed at surfaces of Cu clusters sufficiently large to preclude size effects on reactivity, thus confirming the monofunctional nature of both condensation and esterification pathways. Cu metal is the prevalent phase during catalysis of propanol propanal H 2 mixtures at the conditions of our study. Rates of reduction during H 2 treatment showed that Cu oxide precursors convert to Cu metal at temperatures below 500 K. These samples were treated in H 2 at 553 K before catalytic measurements and used to convert reactant mixtures under reducing conditions that preserve Cu clusters in their metallic state based on thermodynamic considerations and the H 2 /H 2 O ratios prevalent during catalysis. The results presented herein for the condensation of alkanols and alkanals demonstrate the ability of Cu metal clusters to form new C C bonds and to remove O-atoms using C-atoms present within reactants (via decarbonylation or decarboxylation routes), thus preserving valuable H-atoms in the larger oxygenates formed. Concurrent water-gas sift reactions preclude the unequivocal identification of CO or CO 2, formed via decarbonylation or decarboxylation respectively, as the main coproduct of 3-pentanone synthesis Effects of Propanal, Propanol and H 2 Pressure on Condensation and Esterification Rates. The effects of reactant pressures on propanol propanal H 2 reaction rates and selectivities were examined at 503 K on 10% wt. Cu/SiO 2 samples (5.5% dispersion) diluted with SiO 2 to avoid any potential temperature or concentration gradients. Neither pellet diameter nor extent of dilution influenced measured rates or selectivities, thus ensuring that these rates reflect those of the chemical reactions without corruptions by transport artifacts. The only dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

8 Scheme 4. Propanol Propanal Equilibration on Cu-based Catalysts Figure 6. Formation rate for esterification (squares) and aldol condensation (circles) reactions from propanol propanal H 2 mixtures on 10% wt. Cu/SiO 2 (5.5% dispersion) as a function of propanal pressures (filled symbols) or P propanol 3 (P H 2 3 K eq 1 ) 1 (open symbols); K eq 1 : equilibrium constant for propanal propanol interconversion at 503 K. 28 Symbols: experimental data. products detected at all conditions were propyl propionate (ester), 3-pentanone, 2-methyl-3-pentanone, 3-methylpentanal (aldoltype condensation products), CO, CO 2,andsmallamounts of 3-pentanol, 2-methyl-3-pentanol, 3-methylpentanol, propene, and propane; the combined selectivities to the latter two products were <6%. Figure 6 shows esterification and condensation rates (extrapolated to zero residence time to avoid any conversion effects on rates or selectivities) as a function of the measured propanal pressures (bottom abscissa axis) or calculated from propanol and hydrogen pressures assuming thermodynamic equilibration [propanol pressure 3 (H 2 pressure 3 K eq 1 ) 1 ] (top abscissa axis) on 10% wt. Cu/SiO 2. These two curves are identical, consistent with the equilibrated nature of propanal propanol interconversion. Propanol and propanal interconvert on Cu surfaces via adsorbed intermediates with an extent of dehydrogenation that depends on the H* coverage, set in turn via equilibrated H 2 dissociation steps. These steps are shown in Scheme 4 and involve the activation of O H bond in propanol to form surface alkoxides (CH 3 CH 2 CH 2 O*), in a reaction that occurs on Cu(110) and Cu(111) surfaces at even subambient temperatures The resulting alkoxides readily dehydrogenate to alkanals at low temperatures (<370 K), consistent with the fast propanol propanal equilibration at all conditions in our study. Both condensation and esterification are thought to proceed by reactions of adsorbed alkanals with surface alkoxides (C 3 H 7 O*) formed either by O H dissociation in alkanols or partial hydrogenation of alkanals. The effects of H 2 and propanal (or propanol) pressures on condensation and esterification rates (shown below) suggest that propanal is the most abundant surface intermediate (MASI) and that it forms the intermediates required for C O and C C bond formation. Theoretical treatments support the basic character of adsorbed alkoxides and their tendency to preferentially activate acidic H-atoms at C α -H bonds to form adsorbed enolates (CH 3 CH* CHdO). 48 The activation of the propanal by Cu metal surfaces, however, preferentially occurs at the carbonyl C H bond to form a CH 3 CH 2 C*dO* acyl intermediate with an activation energy of 97 kj mol 1, which is 40 kj mol 1 higher than the barrier required for C α H bond activation by adsorbed propoxides. Scheme 5 shows the routes by which CH 3 CH 2 C*dO* and CH 3 CH* CHdO form C O and C C bonds in catalytic sequences that ultimately form esterification and condensation products. Esterification is thought to proceed via nucleophilic attack by the basic oxygen in an adsorbed alkoxide (CH 3 CH 2 CH 2 O*) or acyl (CH 3 CH 2 C*dO*) intermediate at the electrophilic C-atom in the carbonyl group of the adsorbed alkanal (CH 3 CH 2 CHdO*). The resulting C 6 H 13 O 2 * and C 6 H 11 O 2 * intermediates dehydrogenate or hydrogenate, respectively, to form the corresponding ester, as in the accepted mechanism for basecatalyzed esterification. 35,39 Similar esterification paths have been reported for stoichiometric reactions between adsorbed alkoxides and alkanals on Au surfaces containing preadsorbed O* species and for catalytic reactions on nanoporous Au structures with residual Ag also in the presence of O 2. 46,47,49,50 These esterification reactions are examples of general nucleophilic substitutions involving reactions between coadsorbed electrophilic oxygenates and unsaturated hydrocarbons and carbonyls on metals, such as the addition of surface acetate, atomic oxygen or hydroxyl species to alkenes and alkanals, which form vinyl esters, 51,52 epoxides 53 and acids. 54 These bimolecular reactions between propanal and CH 3 CH 2 CH 2 O* or CH 3 CH 2 C*dO* on Cu also resemble steps proposed for Tishchenko reactions of benzaldehyde on alkaline earth oxides, 55 in which vicinal alkanals adsorbed on basic (O 2 ) and acidic cations are involved in rate-determining nucleophilic addition steps. In contrast, C C bonds form via nucleophilic addition of the bound α-carbon of the CH 3 CH* CHdO enolate to the adsorbed carbonyl of the alkanal (CH 3 CH 2 CHdO*) to form HOCH 2 CH(R 1 ) C(R 2 )H O* alkoxides that dehydrogenate to the β-hydroxy aldol products (3-keto-, 2-methyl pentanol) and tautomerize to 3-hydroxy-2-methyl-pentanal. Neither aldol species are detected because of their slow desorption and fast decarbonylation or decarboxylation to form 3- pentanone or their dehydration-hydrogenation to form 2-methyl 3-pentanone and 2-methyl pentanal. Our results show that CdO or CO 2 removal reactions occur faster than dehydration-hydrogenation on Cu surfaces, leading to the preferential formation of 3-pentanone via monofunctional pathways on Cu clusters. Plausible pathways for esterification and condensation reactions are shown in Schemes 6 and 7. The similar effects of reactant concentrations on esterification and condensation rates (Figures 6 and 7) suggest that the pathways involve kinetically relevant steps with intermediates of similar extents of dehydrogenation and that the same kinetic rate equation, albeit with dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

9 Scheme 5. Possible Mechanism for Esterification and Condensation Reactions from Propanal Propanol Mixtures on Cu Catalysts Scheme 6. Mechanism Proposed for 3-Pentanone Formation from Propanol Propanal Intermediate on Cu Catalysts Scheme 7. Mechanisms Proposed for Propyl Propionate Formation from Propanol Propanal Intermediate on Cu Catalysts different kinetic constants, accurately describes the rates of both reactions. The formation of 3-pentanone (Scheme 6) and propyl propionate (Path 1 Scheme 7) involves the activation of the C α -H bond in adsorbed propanal by a vacant site (*) (Step 6, Case i), by another adsorbed propanal (Step 6 0, Case ii), or by vicinal C 3 H 7 O* species (Step 6 00, Case iii) (Schemes 6 and 7) to form C 3 H 5 O* intermediates, which subsequently react with adsorbed propanal in kinetically relevant C C and C O bond formation steps. More specifically, the rate equation for the formation of propyl propionate, which involves the reaction of adsorbed propanal with a surface propoxide (Path 2, Scheme 7; Case iv), is equivalent in functional form to the rate equation for condensation, which reflects hydrogen abstraction from adsorbed propanal by a surface propoxide, because both reactions and rate equations involve reactions between the same two intermediates (CH 3 CH 2 CH 2 O*+CH 3 CH 2 CHdO*), albeit to form different products, in their respective kinetically relevant steps. C 3 H 5 O* is used here to denote CH 3 CH 2 CH 2 C*dO* or CH 3 CH 2 CH* CHO intermediates, because rate data and their mechanistic analysis cannot discern the location of the C H bond activation event and of the subsequent attachment of its products to the surface. Each of the four possible cases discussed above gives a different rate equation, the functional form of which must describe both esterification and condensation rates, in light of their similar measured kinetic behavior. The rate equations for Cases i iii, involving propanal reactions with C 3 H 5 O*, each have two limiting forms, corresponding to the extreme cases of quasi-equilibrated or irreversible H-abstraction steps. The irreversible condensation and esterification reactions in dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,

10 Case iv are different from those in Case iii, but the resulting kinetic expressions are the same, because the kinetically relevant steps for the C O andc C reactions in Cases iii and iv involve C 3 H 6 O* reactions with C 3 H 7 O*, as discussed above. The kinetic and adsorption constants for Case iv, however, differ from those in Case iii, because of the different products that form in these two cases (Table 4). The seven resulting rate equations (derived in Section S2, Supporting Information) are: (i.1.) Quasi-equilibrated Step 6 ½C 3 H 6 OŠ 2 ν ¼ A ½H 2Š 0:5 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð1þ (i.2) Irreversible Step 6 ½C 3 H 6 OŠ ν ¼ A 2 3 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð2þ (ii.1) Quasi-equilibrated Step 6 0 ½C 3 H 6 OŠ 2 ν ¼ A ½H 2Š 0:5 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð3þ (ii.2) Irreversible Step 6 0 ½C 3 H 6 OŠ 2 ν ¼ A 4 3 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð4þ (iii.1) Quasi-equilibrated Step 6 00 ½C 3 H 6 OŠ 2 ν ¼ A ½H 2Š 0:5 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð5þ (iii.2) Irreversible Step 6 00 ½C 3 H 6 OŠ 2 ν ¼ A ½H 2Š 0:5 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð6þ (iv) Irreversible Step 7 0 ½C 3 H 6 OŠ 2 ν ¼ A ½H 2Š 0:5 ð1 þ B 3 ½C 3 H 6 OŠþC 3 ½C 3 H 6 OŠ 3 ½H 2 Š 0:5 þ D 3 ½C 3 H 6 OŠ 3 ½H 2 ŠþE 3 ½H 2 Š 0:5 Þ 2 ð7þ Figure 7. Condensation/esterification ratio for propanol propanal H 2 mixtures on 10% wt. Cu/SiO 2 (5.5% dispersion) as a function of propanal pressures (filled symbols) or P propanol 3 (P H 2 3 K eq 1 ) 1 (open symbols); K eq 1 : equilibrium constant for propanal propanol interconversion at 503 K. 28. The parameters, A i (i =1 7 for 3-pentanone formation) and A 0 i (i =1 7 for propyl propionate formation) are defined in Table 4 and contain kinetic and adsorption constants for elementary steps. The terms B, C, D and E in each of these expressions refer to the adsorption and reaction equilibrium constants that determine the surface coverages of propanal (1/K 4 ), propoxide (K 1 K 2 K /K d ), and hydrogen atoms (1/ K ) and reflect the relative coverages of species at surfaces that are common to both esterification and condensation reactions. The regressed values for B and C are presented in Table 5 for Cases i.2, ii.2 and iii.3). An initial regression analysis of these equations showed that the denominator term for all cases does not contain detectable contributions from the terms for C 3 H 7 OH* (D[C 3 H 6 O][H 2 ]) and H* (E[H 2 ] 0.5 ). Therefore, eqs 1 7 were ultimately regressed without these terms for both condensation and esterification rate data. All irreversible Cases (i, ii, iii, and iv) provided acceptable fits to the rate data (Supporting Information) suggesting that we cannot distinguish based on these data whether C 3 H 6 O* is activated by *, C 3 H 6 O*, C 3 H 5 O* or C 3 H 7 O* surface intermediates. The regressed kinetic parameters for these four dx.doi.org/ /ja207551f J. Am. Chem. Soc. 2011, 133,